Association of white matter hyperintensities with migraine features and prognosis

Migraine is a chronic debilitating headache characterized by recurrent moderate-to-severe headache attacks and autonomic nervous system related-symptoms [1]. Globally, migraine affects approximately 15% of the general population, and it preferentially affects females [2]. Migraine can be regarded as a risk factor associated with white matter hyperintensities (WMHs) [3], which are hyper-intense brain lesions in T2-weighted and Fluid-Attenuated Inversion Recovery (FLAIR) images [4]. Accumulating evidence documented the high incidence of WMHs in patients with migraine [3, 5]. However, the exact correlation between WMHs and the clinical features of migraine remain unclear. A population-based CAMERA study suggested the increased risk of WMHs in migraine patients with higher attack frequencies (≥1 attack per month), compared with patients with lower attack frequencies (< 1 attack per month) [3]. Trauninger et al. demonstrated that both disease duration and attack frequency were associated with WMHs in migraine patients [6]. On the other hand, Toghae et al. observed an association between WMHs and the age and migraine duration of patients, but not with attack frequency [7]. Recent studies have reported that WMHs are associated with the age of patients, and not with disease burden (disease duration and attack frequency) [8, 9]. Therefore, additional studies are required to gain insight into the exact correlation between WMHs and the features of migraine.

Larger or confluent WMHs are usually observed in cerebrovascular diseases and cognitive decline cases [10, 11]. However, WMHs associated with migraine tend to be punctate and mild [10]. Nevertheless, studies that have investigated the clinical implication of WMHs are scarce. Longitudinal population-based studies have previously indicated that WMHs in migraine are not associated with stroke or the decline in cognitive function [10, 12]. However, migraine has a variable short- or long-term prognosis. Some patients achieve complete or partial remission, while others experience persistent or even progressive attacks [13]. Recently, Eggers proposed that WMHs might be caused by multiple microemboli, which are induced by platelet aggregation abnormalities usually observed in migraine patients [14]. This supports the notion that WMHs may reflect an abnormal internal environment in patients. Therefore, the present study aimed to investigate the prognostic value of WMHs in migraine patients. To this end, the T1- and T2-weighted, as well as the T2-weighted FLAIR, magnetic resonance imaging (MRI) data obtained from migraine patients were analyzed to describe the imaging characteristics of WMHs. Furthermore, the association between WMHs and the clinical features of migraine were investigated, and the relationship between WMHs and migraine prognosis were examined.

Migraine is a well-documented risk factor for WMHs [3, 5, 17, 18]. However, to date, the clinical significance of WMHs in migraine prognosis remains unclear. Therefore, in the present study, we explored the association between WMHs and migraine prognosis. Results demonstrated that the presence and degree of WMHs can be associated with unfavorable migraine prognosis. In addition, in the present study, we described the features of WMHs in migraine patients, and observed that WMHs were positively correlated with old age.

WMHs are commonly associated with physiological conditions such as aging and pathological conditions associated with vascular risks such as hypertension [19]. Pathologically, WMHs can result from local brain ischemia at the microvascular level [14, 19]. Several reports have demonstrated a higher incidence of WMHs in patients with migraine, compared to healthy control subjects [3, 5]. In the same context, Schurks et al. demonstrated that migraine is a definite risk of stroke, especially in young women [20]. This suggests the role of ischemia in the mechanism of WMHs in association with migraine. To date, the exact pathophysiology of WMHs is not well-understood. Accumulating evidence revealed that migraine patients may have abnormal platelet activation, impaired endothelial function and hypercoagulability [21–23], which can be potential causes for the development of WMHs. These abnormal vascular conditions might favor the persistence or even the progression of migraine. Consequently, it is reasonable to speculate that WMHs can be correlated with unfavorable migraine prognosis. Indeed, results obtained from the present study demonstrate that both the degree and frequency of WMHs were positively correlated with unfavorable migraine prognosis. To the best of our knowledge, this is the first report that demonstrated the prognostic value of WMHs in migraine patients.

Aging is another important risk factor of the development of WMHs [4, 19]. Among the examined clinical variables, only age and disease duration were correlated to WMHs in our patient cohort. However, a moderate positive correlation was observed between age and disease duration. The disease duration of patients with WMHs was not higher than that of patients without WMHs within the same age brackets. These results imply that the association between disease duration and WMHs could result from the confounding effect of age. Furthermore, we investigated whether age had a confounding effect on the association of WMHs and migraine prognosis. There was no significant difference in age between the improved and non-improved groups. Regression analysis showed WMHs was the independent risk factor for the non-improved outcome with the control of the cofounding effect of age. Finally, the rate of prophylactic treatment was also comparable between these two groups. Collectively, these results indicate that the association between WMHs and migraine prognosis were not affected by age or medications in our patient cohort. However, it is recognized that the prevalence of migraine increases with age from childhood to adulthood, and it peaks at 35 to 39 years of age, after which it gradually decreases, particularly among women after menopause [3]. Meanwhile, WMHs are not static, and in most cases WMHs progress with aging [24, 25]. Therefore, for older migraine patients, WMHs will not serve as a reliable marker for prognosis. It is worth mentioning that a 3-year longitudinal follow-up study revealed a non-significant increase in the number of WMHs in 19.5% of the patient cohort [26]. These results suggest that WMHs would not significantly progress within a relatively short period (3 years). In the present study, patients were re-evaluated after 2–4 years (mean: 3 years). Therefore, this might indicate that the validity of WMHs in migraine prognosis is at least applicable over a relatively short interval (3-year window). Future research should examine its validity through long-term follow ups.

In this study, our results demonstrated that migraineurs with favorable outcome had a higher incidence of aura. Studies analyzing the relationship between the presence of aura and migraine prognosis are scarce. Dahlof et al. observed that aura was associated with poor migraine prognosis in females, but a similar relationship was not observed in male patients [27]. A 5-year follow-up study that investigated the outcome of migraine in children and adolescents failed to find a significant difference between migraine with and without aura, although the percentage of subjects who were free from migraine at follow-up was 30.6% in the case of migraine with aura and 20.3% in the case of migraine without aura [28]. In our study cohort, patients with aura were significantly younger than those without aura. Disease duration and attack duration were significantly lower in patients with aura than those without aura. Patients with aura had a lower incidence of WMHs than those without aura although the difference was not significant. On the other hand, Gozke et al. previously suggested a higher incidence of WMHs in migraine with aura [29]. Moreover, it is well recognized that migraine with aura is a risk associated with ischemic stroke [20]. Therefore, it is plausible to speculate that the better baseline headache condition in patients with aura could contribute to better prognosis compared to patients without aura. However, the exact reason remains to be clarified in future studies.

There was no significant difference of SDS scores between WMHs group and Non-WMHs group. The relationship between depression and WMHs remains unclear [30, 31]. A meta-analysis showed a significant weak association between WMHs and depression (OR: 1.02~ 1.22) [32]. The burden of migraine can be assessed by disease duration, attack frequency, attack duration and headache intensity (VAS) [18]. In the present study, we investigated the impact of WMHs on disease burden. It was observed that WMHs were significantly associated with longer disease duration, while a significant correlation with attack frequency or its duration was not observed. To date, the association between WMHs and migraine features remain controversial [6, 18, 33, 34]. Earlier reports have consistently revealed that WMHs were not associated with migraine features, including disease duration or attack duration [35–37]. On the other hand, Gozke et al. demonstrated that WMHs were associated with a higher frequency of longer disease duration and higher attack frequencies [29]. Similarly, the CAMERA study supported the same conclusions [3]. However, the population of the CAMERA study had high proportions of vascular risk factors (32–42% prevalence of hypertension, 60–66% prevalence of smoking, and 19–29% prevalence of oral contraceptive use), which may lead to confounding bias. This discrepancy might be attributed to the accuracy of the MRI techniques, especially the blurring artifact. Other factors include the demographic characteristics of the study population or discrepancies in the study design [3, 6, 29, 34, 35]. In addition, the measurements of disease burden are usually not stable; that is, disease duration was always affected by age, and attack frequency and attack duration often show changeable patterns over time; while the VAS had strong subjectivity. Therefore, future research should focus on more stable parameters to assess the disease burden of migraine.

Taken together, the results obtained from the study suggest that WMHs may predict unfavorable migraine prognosis. Therefore, our results could lead to the alteration of the treatment protocol for migraineurs with WMHs. That is, physicians could apply more positive treatment strategies to achieve a more favorable prognosis in patients with high WMHs scores. Furthermore, our results also indicate that WMHs have a closer association with age than the clinical features of migraine.

Nevertheless, the present study had a few limitations. The relatively small number of enrolled patients is considered to be the main limitation. Furthermore, the absence of a control group precluded definitive conclusions about the nature of the observed alterations in WMHs or whether their degree is beyond normal aging. Age should be controlled in the design of the study. Thus, future work should focus on investigating the implication of WMHs among relatively young migraine patients. Similarly, the heterogeneity of the patient cohort such as migraine with and without aura, episodic migraine and chronic migraine, should be improved. Different migraine types possibly have different effects on the prognosis. The relatively low migraine frequency at baseline in our study is also a major limitation as prognostic information may be of greater value in high frequency migraine states. However, the small sample size limited the stratified analysis of frequency in the case of controlling the effect of age. The method for migraine prognosis categorization was one-sided. It reflected the change of frequency but it did not investigate the current frequency level, headache intensity or even response to acute therapy. However, there is no standard prognostic categorization for migraine yet. Future prospective multicenter studies with more controlled conditions (migraine type and age) and long-term follow up should be conducted to confirm these results. Future studies should also employ more stable parameters that assess disease burden, in order to further confirm the clinical significance of WMHs. It is worth mentioning that in the present study, we could not definitively investigate other WMH-associated risk factors, including hypertension, diabetes, hyperlipidemia, hyperhomocysteinemia, hyperuricemia, hypercoagulability, heart diseases, kidney diseases, inflammation and autoimmune diseases. These conditions may impact the strength of our conclusions regarding the nature of WMHs and their effect on migraine. Future studies should be controlled for the confounding effects of the above mentioned conditions.

This study suggests that WMHs can predict short-term unfavorable migraine prognosis, thereby providing a new insight into the clinical significance of WMHs in migraine. Meanwhile, it demonstrates that WMHs in migraine patients are generally mild, mostly located in the frontal and parietal lobes, and may have a closer association with age than headache characteristics.